Anisotropic Relative Permeabilities for Characterising Heavy-Oil Depletion Experiment

Abstract In recent years, the use of pore-scale network models has greatly advanced our understanding of solution gas drive processes by accounting for the complex dynamics operating at the microscopic scale. Moreover, it has also been demonstrated that a pore-network model, when suitably anchored to core material is able to provide both qualitative and quantitative descriptions of relative permeability and hydrocarbon recovery. In contrast, many so-called "experimental" depletion drive relative permeabilities are not measured directly but are generally obtained by history-matching laboratory production data with reservoir simulators, often resulting in very low gas relative permeabilities that are difficult to explain from a physical viewpoint. Although pore-scale network models have been successfully used in the past to match raw production data, the steady-state relative permeabilities calculated from such models commonly predict much slower gas saturation build-up than that found experimentally. Some previous authors have related this low gas saturation build-up to the difference in the definition of critical gas saturation between reservoir simulators and pore-network models. However, the dentritic nature of gas-cluster topology in network models, especially in the presence of other forces, such as gravity or strong viscous pressure gradients, clearly suggests that significant anisotropy may exist in relative permeability due to the balance of forces at this scale. In the present work, we describe how the naïve process of scaling up steady-state relative permeabilities obtained from pore-scale network models to the laboratory scale may contribute significantly to the difficulty in history-matching experimental production. By considering the influence of the various forces (capillary, gravity, viscous) on the topology of the growing gas clusters and by accurately incorporating anisotropic network-model relative permeabilities, we show that high gas saturation build-up, consistent with experimental observations can be obtained from reservoir simulation.